1 Object Detection Using Faster R-CNN

[@shubhamaditya7481]

Please never stop making videos. The way you explain slowly and step by step is very very very simple to understand and mind it we dont get this regularly

[@CodeWithAarohi]

Glad to know that my videos are helpful. And Sure I will keep on making videos in future.

[@ssshukla26]

This is simply the best explanation for FRCNN, I was confused with RPN, now this has cleared my doubts. Thanks for such a great video.

[@CodeWithAarohi]

Glad my video helped you !

[@harshrawlani6793]

I have my interview next week. I tried to read medium articles/ research paper but didn't understand then I found your videos. And now I am binge watching. Keep up the good work. ☺️

[@CodeWithAarohi]

Glad it was helpful! Best of Luck😊

[@Akshtest]

By far, it is the best explanation of Faster RCNN

[@CodeWithAarohi]

Glad it helped!

[@jayjain6204]

Really amazing, wish we had teachers like you!

[@CodeWithAarohi]

Glad my videos helped you

[@monindrasinha1296]

best video for a research student thank you .......all your object detection videos helps me a lot.

[@CodeWithAarohi]

Glad to hear that. Good luck for your research:)

[@anoopyadav4081]

I follow so many KZbin channels for understanding computer vision but didn't understand fully what they taught. Now m happy that I understood the concept of faster rcnn. Thanks madam for great explanation. Please keep educating us on DL/CV n all.

[@CodeWithAarohi]

Welcome

[@elokamaylia2466]

I really understand with ur explanation. Thank you very much 🤗

[@CodeWithAarohi]

You're welcome 😊

[@ChattyCheerful]

Thankyou aarohi! I feel your teachings are the best. Its a blessing to take your lectures. Much appreciated : )

[@CodeWithAarohi]

Glad to hear that 😊

[@himanshumangoli6708]

In RPN input we have image and for that image we will create anchor box of different size. The portion of image which have some part of our desired image we called this foreground image if the IOU score is greater than 0.5 then we will passed this image to some sort of CNN which will produce some feature maps and these feature map will go to ROI

[@ceesha1]

easy version of Faster R-CNN,very much understandable...Thanks for your videos

[@CodeWithAarohi]

You are welcome!

[@justinamichael316]

Clear and superb explanation. Thank you.

[@CodeWithAarohi]

Glad my video is helpful!

[@akhiljithk7173]

5:40 how do we get the original yellow box for calculating IoU?

[@shaileshyadav5806]

Really amazing, teachers like you!

[@CodeWithAarohi]

Thank you!

[@YAlsadah]

What an amazing explanation. Thank you very much!

[@CodeWithAarohi]

Glad it was helpful 😊

[@rakeshkumarkuwar6053]

Really Glad that, I'm across your channel; such a nice explanation. Thanks for creating such helpful videos.

[@CodeWithAarohi]

Welcome!

[@curious523]

Thank you for the amazing videos. I have a question regarding using VGG. Input accepted by VGG is in the form of (224,224,3) while my train data frame is in 2d form (e.g. (1997, 7)). How can I convert my data frame to be accepted by VGG. I used reshape but got this error: 'ValueError: cannot reshape array of size 13979 into shape (1997,7,3)' . It considers the size of the data frame. Does anyone any idea? thanks in advance

[@CodeWithAarohi]

To use VGG with your dataset, you will need to convert your data to the appropriate format.You mentioned that you tried to reshape your data using reshape(), but you received an error. The error suggests that the reshaped array has a different size than the original array. To fix this error, you can try reshaping your data using the following code: import numpy as np # assume X_train is your training dataset X_train_reshaped = np.reshape(X_train, (1997, 7, 3))

[@bishwarup1429]

Thank you ma'am for these video. You are the next upcoming best youtube channel for Deep Learning. Please continue this good deed. Thank you.

[@CodeWithAarohi]

You are welcome 🙏

[@tahanifennir9796]

You are amazing thank you for the good explanation

[@CodeWithAarohi]

Glad it was helpful!

[@abhijeetkhandwe7909]

Finally understood Faster RCNN

[@CodeWithAarohi]

Glad my video helped you

[@MahsaPourhosein]

It was perfect and very helpful.

[@CodeWithAarohi]

Glad it was helpful!

[@surflaweb]

not lots of words, concise and clear.. thanks so much

[@CodeWithAarohi]

Welcome

[@ajitkumar15]

really great video with simplicity and best explanation..Thank you so much

[@CodeWithAarohi]

Welcome

[@aditi1706]

Hi Aarohi, thanks for the wonderful contains. i am building custom model for detecting small scratches and dents ( in micron). can you suggest me the algo i should use?

[@CodeWithAarohi]

When it comes to detecting small scratches and dents in micron level, the accuracy and precision of the object detection algorithm are crucial. The choice of algorithm depends on several factors, such as the size of the objects, the number of objects in the image, and the available hardware resources. In general, two-stage object detection algorithms like Faster R-CNN and Mask R-CNN tend to be more accurate but slower, while one-stage object detection algorithms like RetinaNet and YOLO tend to be faster but less accurate.

[@geo6086]

First of all, thank you very much for your tutorial. If I want to use several models in the video to detect targets in remote sensing images, how should I do it? I sincerely hope that you will make a tutorial.

[@CodeWithAarohi]

Will try to do a video on after finishing my pipelined videos

[@geo6086]

@@CodeWithAarohi Thank you very much for your conscience production

[@Maver1ck3981]

simple and quick to understand.Great!

[@CodeWithAarohi]

Glad it was helpful!

[@sml9360]

You are really amazing. The way you explained is simply superb without any confusion. God bless you and your family, Keep rocking. U deserve a big applause. One small thing noticed, audio was fluctuating somewhere in the middle. Make sure audio is working properly. Thank you so much. Aarohi.

[@CodeWithAarohi]

Thankyou for appreciating my work and thankyou for letting me know this audio issue. I will see to it.

[@asim-gandu-phenchod]

Nice explanation. Keep it up

[@CodeWithAarohi]

Thanks

[@rudrapratap8270]

Amazing explanation!!!

[@CodeWithAarohi]

Glad you think so!

[@priyabratapanda1216]

Nice Aarohi really intuitive!

[@CodeWithAarohi]

Thankyou

[@kiruthikasubramaniam5425]

Thank you so much mam.awesome work

[@CodeWithAarohi]

Welcome

[@siddhantsarraf]

Thanks for such a wonderful explanation 🙏🏼🙏🏼

[@CodeWithAarohi]

Welcome

[@taranpreetsingh6666]

which technique you have used for image annotation?

[@darshandhobi3133]

Thank u so much mam it is very helpful

[@CodeWithAarohi]

Welcome

[@atulbharadwaj774]

Will IoU be calculated based on comparison with ground truth?

[@roboticsforunitedindia9421]

I want a complete yolo course means yolo v1,yolo,2,YOLOv3 ,yolov4, yolov5,yolor course ,you have online course mam

[@CodeWithAarohi]

No, I don’t have any online course

[@ranjan561]

can we use Faster RCNN for numerical data ?

[@CodeWithAarohi]

While Faster R-CNN was originally designed for object detection in images, it can be used for other applications as well, including numerical data. However, using Faster R-CNN for numerical data may not be the best choice in most cases, as Faster R-CNN is optimized for object detection in images and may not perform well on numerical data.

[@aryan_01793]

Ma'am I have question that when it extract the feature maps after getting the proper Anchor Boxes does the CNN applied here completely or just some part of CNN to get the feature maps?

[@CodeWithAarohi]

Some part of cnn to extract features basically leave the fully connected layers of the vgg and the last cnn layer before fc layers . Pick feature Map from there

[@aryan_01793]

​@@CodeWithAarohi Got it. Thank you ma'am.

[@ismailkadkc1121]

Thank you for great video :) I would like to ask which tensorflow model is best to detect fruits on drone images. Drone hight almost 20 meters?

[@techsavy5669]

I am working on a similar problem with an hd drone at 100 meters. Did you figure out any solution yet! I emailed Aarohi also, i think she is thinking! Thanks :)

[@kaustuvsarangi8788]

How the Faster RCNN using my collection image

[@latabai3533]

Which algorithm is best for traffic sign recognition???

[@CodeWithAarohi]

You can use Convolutional neural Networks like SSD, Yolo, FasterRcnn

[@latabai3533]

@@CodeWithAarohi please make full video on traffic sign recognition

[@CodeWithAarohi]

@@latabai3533 Sure Will do that soon

[@latabai3533]

As soon as possible and also difference between fast cnn and cnn.

[@furkankaragoz8196]

Hi madam, Firstly i appreciate for your videos, I have a question related with the faster r-cnn network, which i am struggled. - as default parameter on faster r-cnn config file image resizer is 600x1000. which means during the training your images will resize that specific size, right? - After traning stage i will have weights and i will use this weights file and try to predict an image. -My question is : During the prediction what if my images is 3000x3000. what will happen? The faster r-cnn network resizes it itself? This makes me so complicated. Sorry if i made a mistake Thanks in advance.

[@CodeWithAarohi]

Thankyou for appreciating my work! You need to resize the image, convert it into array, use other pre processing techniques which fasterrcnn used for image scaling yourself. And then predict

[@sadekasany949]

Thank you so much for this tutarial.

[@CodeWithAarohi]

welcome

[@autonomousselfdrivingnonco8334]

Awesome series maam, can u also share the paper u referred for the faster rcnn

[@CodeWithAarohi]

will share the link soon

[@roboticsforunitedindia9421]

I want join your complete yolo course

[@hadermeche5401]

thank you very much 🙏

[@CodeWithAarohi]

Welcome!

[@taniasultana5865]

Thanks for your work. If (Normal MRI) = 500 and (MRI with tumor) = 90 then how can I detect the tumor using faster RCNN? Because, in case of Normal MRI no need any bounding box.

[@CodeWithAarohi]

just train your network for MRI (with tumor) class. So that it can detect if there is a tumor. And if it is not detecting anything on MRI that itself means there is no tumor in that particular MRI.

[@taniasultana5865]

@@CodeWithAarohi thank you

[@AnthonyGarland]

You could really use a better mic. It is hard to understand you when your voice is fading in and out.

[@CodeWithAarohi]

Thankyou for the feedback but already implemented your suggestion

[@MEET9426]

CN you please clear the audio...at some places it's added noise

[@CodeWithAarohi]

I am so sorry for inconvenience. I will take care of this from my next videos. And thank you for letting me know.

[@foxoism]

You’re amazing

[@CodeWithAarohi]

Thankyou

[@atanukundu9062]

Mam can you share the ppt also??

[@anuragshrivastava7855]

In rcnn for a single image it propose 2000 proposals

[@hoaniam1394]

hi i found your post awesome!!! Can I have the PowerPoint file please?

[@sadekasany949]

Ma'am can u give me mask rcnn tutarial link?

[@CodeWithAarohi]

I hadn't made that tutorial yet. Will make the video soon and share

[@ajithkumarspartan]

Could you do similar video for Mask rcnn also, this is so good

[@CodeWithAarohi]

Yes of course!

[@sreedeviprasad7082]

Good class but Sound quality is poor

[@CodeWithAarohi]

Sorry for inconvenience.

[@thepresistence5935]

marvelous

[@CodeWithAarohi]

Thankyou!

[@amurumanasa3135]

Can you send the object detection code using vgg16

[@CodeWithAarohi]

vgg16 is not for object detection. Vgg-16 is used for feature extraction

[@krishnaponnuru3973]

Thnak You....!!!!!

[@CodeWithAarohi]

welcome 😊

[@AHMADKELIX]

permission to learn mom.thanks

[@CodeWithAarohi]

Absolutely!

[@AHMADKELIX]

@@CodeWithAarohihow to know accuracy result or mAP using this code ?

[@vibhu613]

great

[@CodeWithAarohi]

Thanks!

[@luiscao7241]

Thanks

[@CodeWithAarohi]

Welcome

[@luiscao7241]

@@CodeWithAarohi you have done a series of excellent video. I am very happy to have little support by joining your channel. We should have some short talk on it. Thanks

[@CodeWithAarohi]

@@luiscao7241 thankyou for your support

[@abhishekkumarpandey1862]

Subscribed!

[@CodeWithAarohi]

Thanks

[@squallx1142]

英文本就不好，还要听方言。太苦了。

[@SS-yb1qd]

I love u and ur explaination. I work in Germany currently. Will u marry me?

[@MohsenRezaei-r6o]

good , but not complete.